Get ValidationPackage over network #727
Conversation
…age() (and initialized()) from get_validation_package reducer
…f the entry is an AgendId entry.
Those were mixed up
Codecov Report
@@ Coverage Diff @@
## develop #727 +/- ##
===========================================
+ Coverage 72.79% 73.31% +0.53%
===========================================
Files 129 138 +9
Lines 4541 4690 +149
===========================================
+ Hits 3305 3438 +133
- Misses 1236 1252 +16
Continue to review full report at Codecov.
|
| dna_hash: network_state.dna_hash.clone().unwrap(), | ||
| agent_id: network_state.agent_id.clone().unwrap(), | ||
| address: entry.address().to_string(), | ||
| content: serde_json::from_str(&serde_json::to_string(&entry_with_header).unwrap()) |
There was a problem hiding this comment.
We can and we should. But I didn’t want to change the net crates in this PR and instead just use it. Follow up.
| agent_id: network_state.agent_id.clone().unwrap(), | ||
| address: link.base().to_string(), | ||
| attribute: String::from("link"), | ||
| content: serde_json::from_str(&serde_json::to_string(&entry_with_header).unwrap()) |
| dna_hash: network_state.dna_hash.clone().unwrap(), | ||
| agent_id: get_dht_data.from_agent_id.clone(), | ||
| address: get_dht_data.address.clone(), | ||
| content: serde_json::from_str(&serde_json::to_string(&maybe_entry).unwrap()).unwrap(), |
core/src/network/state.rs
Outdated
| /// None: process started, but no response yet from the network | ||
| /// Some(Err(_)): there was a problem at some point | ||
| /// Some(Ok(None)): no problem but also no entry -> it does not exist | ||
| /// Some(Ok(Some(entry))): we have it | ||
| type GetEntryResult = Option<Result<Option<Entry>, HolochainError>>; |
| ) { | ||
| let maybe_validation_package = match get_entry(&requested_entry_address, &context) { | ||
| Ok(entry) => await!(build_validation_package(&entry, &context)).ok(), | ||
| Err(_) => None, |
There was a problem hiding this comment.
Do we really send a None value back across the network if there was an error? I have the feeling that we at least have to tell the requester that we had some kind of error... followup?
| fn address(&self) -> Address { | ||
| match &self { | ||
| Entry::AgentId(agent_id) => agent_id.address(), | ||
| _ => Address::encode_from_str(&String::from(self.content()), Hash::SHA2256), |
There was a problem hiding this comment.
hmm, this seems like it should always defer to the .address() method of the inner value?
There was a problem hiding this comment.
Are you sure? Wouldn't that mean that the entry type is not included in the hash? In Montreal we agreed that the "content" that gets hashed needs to include the type.
|
@thedavidmeister raw serde handling because |
|
@lucksus right... i keep leaving the same feedback for the same thing 😅 carry on then |
Context
Implements this state: https://realtimeboard.com/app/board/o9J_kyiXmFs=/?moveToWidget=3074457346314392714
Implementation steps in this PR
DirectMessagewith variantsRequestValidationPackageandValidationPackageAction::GetValidationPackagewith reducer and action creator that lets the network send aDirectMessage::RequestValidationPackageto the entries sourceworkflows::respond_validation_package_requestthat checks if we have the entry in the source chain, builds the validation package and then triggers sending of the response via:Action::SendDirectMessagethat makes the network send any (initial or responding) node2node messageDirectMessage::ValidationPackagethat updates the state so theGetValidationPackageFuturecan resolveAdditional required changes in here
Entry::address()returns the right address also for agent_id entries which override the hashing of the content and instead have their public key be the address (6f494af and b66068a)ProtocolWrapper::SendResultandProtocolWrapper::HandleSendResultwere mixed up (right, @neonphog? Or did I get it wrong?) I've swapped them to have handle always mean that the network requests core to handle something. (76fa842 and e812ae7)